Integrating effort- and gradient-based approaches in optimal design of experimental campaigns

Model-based design of optimal experimental campaigns comprising multiple parallel runs can prove computationally challenging. Effort-based methods can help in overcoming some of these challenges through discretising the experimental design space. However, the quality of the resulting approximate solutions depends heavily on this a priori discretisation. This paper presents a methodology for integrating the appealing features of effort-based methods with those of conventional gradient-based approaches, with a view to computing maximally-informative campaigns of experiments for improving parameter precision. The effectiveness of the methodology is demonstrated on a case study involving a microbial culture dynamic model.